AI server has excellent graphics processing ability and high-performance computing ability. Compared with ordinary servers, there is no difference in memory, storage and network. It mainly needs more internal and external memory in big data, cloud computing and artificial intelligence to meet the needs of data collection and collation.
The three elements of deep learning are data, algorithm and computing power. Among them, data is the foundation, algorithm is the tool, and computing power is the booster. The improvement of computing power promotes the development of deep learning. Before deep learning, the development is slow. In addition to the reason of algorithm, a very important reason is the lack of computing power, The most important support for computing power is AI server (here mainly refers to general AI server, GPU server).
从服务器的硬件架构来看，AI 服务器是采用异构形式的服务器，在异构方式上可以根据应用的范围采用不同的组合方式，如 CPU+GPU、CPU+TPU、CPU+其他的加速卡等。与普通的服务器相比较，在内存、存储、网络方面没有什么差别，主要在是大数据及云计算、人工智能等方面需要更大的内外存，满足各种数据的收集与整理。
In terms of the hardware architecture of the server, AI server is a heterogeneous server. In terms of heterogeneous mode, different combinations can be adopted according to the application scope, such as CPU + GPU, CPU + TPU, CPU + other acceleration cards, etc. Compared with ordinary servers, there is no difference in memory, storage and network, mainly in big data, cloud computing, artificial intelligence and other aspects, which need more internal and external memory to meet the needs of various data collection and collation.
我们都知道普通的服务器是以 CPU 为算力的提供者，采用的是串行架构，在逻辑计算、浮点型计算等方面很擅长。因为在进行逻辑判断时需要大量的分支跳转处理，使得 CPU 的结构复杂，而算力的提升主要依靠堆砌更多的核心数来实现。
We all know that the common server is the provider of computing power based on CPU. It adopts the serial architecture and is good at logic computing and floating-point computing. Because a lot of branch jump processing is needed in logic judgment, the structure of CPU is complex, and the improvement of computing power mainly depends on stacking more cores.
但是在大数据、云计算、人工智能及物联网等网络技术的应用，充斥在互联网中的数据呈现几何倍数的增长，这对以 CPU 为主要算力来源的传统服务提出了严重的考验，并且在目前 CPU 的制程工艺、单个 CPU 的核心数已经接近极限，但数据的增加却还在持续，因此必须提升服务器的数据处理能力。因此在这种大环境下，AI 服务器应运而生。
However, with the application of network technologies such as big data, cloud computing, artificial intelligence and Internet of things, the amount of data in the Internet is growing exponentially, which poses a serious challenge to the traditional services with CPU as the main source of computing power. At present, the processing technology of CPU and the number of cores of a single CPU are close to the limit, but the increase of data continues, Therefore, the data processing ability of the server must be improved. Therefore, in this environment, AI server came into being.
现在市面上的 AI 服务器普遍采用 CPU+GPU 的形式，因为 GPU 与 CPU 不同，采用的是并行计算的模式，擅长梳理密集型的数据运算，如图形渲染、机器学习等。在 GPU 上，NVIDIA 具有明显优势，GPU 的单卡核心数能达到近千个，如配置 16 颗 NVIDIA Tesla V100 Tensor Core 32GB GPUs 的核心数可过 10240 个，计算性能高达每秒 2 千万亿次。且经过市场这些年的发展，也都已经证实 CPU+GPU 的异构服务器在当前环境下确实能有很大的发展空间。
Nowadays, AI servers on the market generally adopt the form of CPU + GPU, because GPU is different from CPU, which adopts parallel computing mode and is good at sorting out intensive data operations, such as graphics rendering, machine learning, etc. On GPU, NVIDIA has obvious advantages. The number of single card cores of GPU can reach nearly 1000. For example, the number of cores with 16 NVIDIA Tesla V100 tensor core 32GB GPUs can exceed 10240, and the computing performance can reach 20 billion times per second. And after years of market development, it has been confirmed that CPU + GPU heterogeneous server in the current environment can really have a lot of development space.
但是不可否认每一个产业从起步到成熟都需要经历很多的风雨，并且在这发展过程中，竞争是一直存在的，并且能推动产业的持续发展。AI 服务器可以说是趋势，也可以说是异军崛起，但是 AI 服务器也还有一条较长的路要走。
But it is undeniable that every industry needs to go through a lot of wind and rain from the beginning to maturity, and in the process of development, competition always exists, and can promote the sustainable development of the industry. AI server can be said to be a trend, it can also be said to be the rise of foreign forces, but AI server also has a long way to go.
What’s the difference between AI server and normal server?
1、从服务器的硬件架构来看，AI 服务器是采用异构形式的服务器，在异构方式上可以根据应用的范围采用不同的组合方式，如 CPU+GPU、CPU+TPU、CPU+其他的加速卡等。与普通的服务器相比较，在内存、存储、网络方面没有什么差别，主要在是大数据及云计算、人工智能等方面需要更大的内外存，满足各种数据的收集与整理。
1. In terms of the hardware architecture of the server, AI server is a heterogeneous server. In terms of heterogeneous mode, different combinations can be adopted according to the application scope, such as CPU + GPU, CPU + TPU, CPU + other acceleration cards, etc. Compared with ordinary servers, there is no difference in memory, storage and network, mainly in big data, cloud computing, artificial intelligence and other aspects, which need more internal and external memory to meet the needs of various data collection and collation.
2、卡的数量不一致：普通的 GPU 服务器一般是单卡或者双卡，AI 服务器需要承担大量的计算，一般配置四块 GPU 卡以上，甚至要搭建 AI 服务器集群。
2. The number of cards is different: ordinary GPU servers are usually single card or dual card, AI servers need to undertake a lot of calculation, generally configure more than four GPU cards, and even need to build Ai server cluster.
3、独特设计：AI 服务器由于有了多个 GPU 卡，需要针对性的对于系统结构、散热、拓扑等做专门的设计，才能满足 AI 服务器长期稳定运行的要求。天 下 数据电话 4006-388-808 官网：www.IdCbest.Com。
3. Unique design: as AI server has multiple GPU cards, it needs special design for system structure, heat dissipation, topology, etc. in order to meet the requirements of long-term stable operation of AI server. Tianxia data telephone 4006-388-808 official website: http://www.idcbest.com.
What are the application scenarios of AI server?
基于 AI 服务器的优势，AI 服务器在医疗、搜索引擎、游戏、电商、金融、安防等行业有着广泛的应用。
Based on the advantages of AI server, AI server is widely used in medical, search engine, game, e-commerce, finance, security and other industries.
1. Medical image intelligent analysis scene: through machine vision, knowledge mapping, deep learning and other artificial intelligence technology, simulate the thinking of medical experts, reasoning, help doctors locate the disease, and assist in making diagnosis.
2. Face recognition, speech recognition, fingerprint recognition scene: through deep learning, machine learning and other technologies, image, video and other image data training can be realized.
3. Security monitoring scene: using knowledge mapping technology and deep learning technology, it can be applied to human body analysis, image analysis, vehicle analysis, behavior analysis and other security scenes.
4. Retail prediction scenario: through machine learning, based on the store’s historical sales data, it can accurately predict the future sales, and provide more accurate decision analysis for operators. In addition, in the retail industry, there are some common application scenarios, such as no one selling, face payment and so on.
5、金融服务场景：通过语音、唇形、表情合成技术和深度学习等技术，克隆出与真实面审员一样 AI 视频面审员，可以准确无误的与客户进行多轮面谈。
5. Financial service scene: through voice, lip shape, expression synthesis technology and deep learning technology, AI video face reviewers are cloned, which are the same as real face reviewers, and can accurately conduct multiple rounds of interviews with customers.